Using cognitive psychology to understand GPT-3

• Published in: Proceedings of the National Academy of Sciences - PNAS Vol. 120; no. 6; p. e2218523120
• Main Authors: Binz, Marcel, Schulz, Eric
• Format: Journal Article
• Language: English
• Published: United States National Academy of Sciences 07.02.2023
• Subjects: Agents (artificial intelligence); Cognitive ability; Cognitive Psychology; Decision Making; Humans; Language; Learning; Perturbation; Physical Sciences; Problem Solving; Reasoning; Reinforcement, Psychology; Signatures; Social Sciences; cognitive psychology; decision-making; artificial intelligence; language models; reasoning
• ISSN: 0027-8424; 1091-6490; 1091-6490
• DOI: 10.1073/pnas.2218523120

We study GPT-3, a recent large language model, using tools from cognitive psychology. More specifically, we assess GPT-3’s decision-making, information search, deliberation, and causal reasoning abilities on a battery of canonical experiments from the literature. We find that much of GPT-3’s behavior is impressive: It solves vignette-based tasks similarly or better than human subjects, is able to make decent decisions from descriptions, outperforms humans in a multiarmed bandit task, and shows signatures of model-based reinforcement learning. Yet, we also find that small perturbations to vignette-based tasks can lead GPT-3 vastly astray, that it shows no signatures of directed exploration, and that it fails miserably in a causal reasoning task. Taken together, these results enrich our understanding of current large language models and pave the way for future investigations using tools from cognitive psychology to study increasingly capable and opaque artificial agents.